Capability Without Revenue
Read full version (8 min)The AI economy is generating spectacular proof of capability and almost no auditable proof of durable revenue. From indie agent builders stacking Mac Minis to OpenAI's 900-million-user engagement problem, every signal points the same direction: the cost of building v1 has collapsed, but the cost of everything after hasn't moved.
The Agent Gold Rush Is Long on Screenshots, Short on Receipts
The agent hype cycle has reached the phase where the loudest success stories are indistinguishable from marketing.1 Tech Twitter is saturated with Mac Mini stacks and dashboard glow as revenue proxy. The documented cases of durable, agent-driven income remain thin. Auto-trading narratives founder on a basic asymmetry: if the inefficiency is obvious enough for a Mac Mini to find, a quant fund found it last Tuesday.
Where agents are generating real revenue is exactly where no one goes viral: back-office automation, compliance documentation, reconciliation workflows. Boring, vertical, sticky. The people making money from the agent boom are selling infrastructure and vertical automation. The shovels are fine. The livestreamed gold rush remains unpredictable.
Coding Agents Multiply What You Already Know
Drew Breunig's framing of coding agents makes two sharp observations. First: skilled developers dramatically underestimate the intuitive knowledge they bring to their prompts. When an experienced engineer says Claude Code "just worked," you're not seeing the domain expertise embedded in a relatively specific prompt. Expertise is invisible to the person who has it.
Second: most of what gets hyped as agentic coding output is personal software, not products. "Code is free, as in puppies." The v1 is trivial. Everything that makes it a product — testing, support, cross-platform, marketing — remains expensive. AI has collapsed the cost of manifestation without reducing the cost of maintenance or market fit.
The Commoditization Problem
Anthropic's Claude Opus 4.6 represents a genuine capability jump. What matters more is the context. Benedict Evans's strategic analysis diagnoses the entire foundation model layer: half a dozen organizations shipping competitive frontier models, leapfrogging each other every few weeks, with no known mechanic for any of them to build a durable lead. No network effects. No self-reinforcing market share.
Evans's browser analogy is the sharpest articulation to date. A chatbot is an input box and an output box. Microsoft won browsers and it turned out not to matter — value accrued to the experiences built on top.
OpenAI has 800-900 million weekly active users and shallow engagement. Eighty percent sent fewer than 1,000 messages in all of 2025. Only 5% pay. If people who know what this is still can't think of something to do with it on an average day, a better model may not be what's missing. The competition is shifting to brand and distribution, which is exactly what happens when the underlying product resists differentiation.
Distillation and the AI Layoff Script
Anthropic accused DeepSeek, Moonshot AI, and MiniMax of industrial-scale distillation — 16 million exchanges through 24,000 fraudulent accounts. The technical reality is more nuanced. DeepSeek's 150,000 exchanges are negligible at training scale. The volume came from MiniMax and Moonshot, focused on agentic reasoning where Claude leads.
The tension Anthropic can't resolve: you cannot simultaneously offer the world's best model as an API product and object when that access trains competitors. Expect other labs to follow with disclosure campaigns. The technical merits will matter less than the utility of these narratives as trade-policy instruments.
Meanwhile, Jack Dorsey's Block memo — job cuts framed as AI-driven restructuring — continues a hardening pattern.2 "We're replacing headcount with AI" is becoming boilerplate. The enterprise version of the Mac Mini stack. AI as corporate narrative device is following the same trajectory as "digital transformation" a decade ago.
What to Watch
The post-demo transition enters its test phase. The next quarter will be defined by retention data and revenue durability, not capability announcements. If OpenAI's engagement numbers don't deepen, the browser analogy becomes prophecy.
Foundation model pricing hits the floor. Commodity dynamics are accelerating. The question isn't whether prices drop but how fast, and what that does to labs dependent on API margins.
Distillation as trade-policy instrument. The gap between technical reality (diminishing returns in an RL-dominated paradigm) and political utility (justifying restrictions) is where actual policy will get made.
Way Enough is written collaboratively by a human and an AI agent.